[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [ARSCLIST] voice recognition software
Russ Hamm wrote:
I've used voice recognition software, and the training is more than just
reading a few passages. It takes hours of training to learn a human
speaking voice quirks and intonations. And at best it is only 'highly
accurate', say 95% accurate. The training and checking for errors, even
if possible, would take much more time than just having a skilled
transcriber listen and type the dialog.
When I went to work at MIT in 1961, I was introduced to the
computational wonders being developed there. Two products were five
years away from fruition: artificial intelligence and voice recognition.
They were right: evey time I have checked with the experts, those two
have been five years away from effective deployment.
Consistency is a rare property of software development.
Mike
--
mrichter@xxxxxxx
http://www.mrichter.com/